Sustainability

We practise what we preach.

A platform built to improve energy efficiency should hold itself to the same standard. Here is what this site and a full analysis actually cost in carbon.

This page visit

Estimated emissions per visit

Data transfer CO₂

Tree offset time

Measured with CO2.js from The Green Web Foundation, Sustainable Web Design model, green hosted.

Per AI analysis

A full property analysis, six images

Image transfer~15 MB
AI compute (GPU)~4.2g CO₂
API overhead~0.8 MB

Estimates based on published inference costs, average image sizes of 2.5 MB, and the Sustainable Web Design transfer model. The GPU figure is derived from published ML energy research, not measured on our own hardware.

The AI layer

How an analysis stays lean

Inference is the most carbon-intensive thing we do, so it is the part we constrain hardest.

01

Right-sized models

Individual image analysis uses the small model. The full model runs once, for final synthesis. That cuts GPU compute by roughly 60% against running the large model on every image.

02

Results cached permanently

Every analysis is stored. Viewing your results a hundred times costs zero additional API calls, and nothing expires and forces a re-computation.

03

User-triggered only

No speculative pre-analysis, no background polling, no auto-refresh. The AI runs when you press the button, so every GPU cycle was asked for.

04

Parallel processing

All six images are analysed concurrently in one batch rather than in sequence, which cuts total wall-clock time and connection overhead.

05

Efficient prompting

Structured output schemas mean the model returns exactly the fields we need. Fewer output tokens is less compute per request.

06

Scoped outputs

Each stage produces only the data its own context needs — component detection, EPC estimation, or report generation — with no redundant re-analysis.

Understanding your home's energy performance costs about 4.6g of CO₂.

Less than boiling a kettle. A full analysis — six images, component detection, EPC estimation — for the carbon cost of a few seconds of browsing.

Everything else

Beyond the analysis

Infrastructure

Vercel edge network on renewable energy. A serverless backend that scales to zero. CDN-cached static assets cutting repeat transfers sharply. Green hosting verified through The Green Web Foundation.

Codebase

Automatic code splitting, CSS that purges unused styles, dynamic imports for heavy libraries, and tree-shaking that keeps dead code out of production bundles.

Data practices

WebP image optimisation. Real-time sync that sends deltas rather than full payloads. Incremental progress tracking, so completed steps are never re-uploaded.

Dependencies

Regular audits to remove unused packages, a preference for lightweight libraries, and zero-dependency utilities wherever a plain function replaces a package.

Performance

Core Web Vitals monitored continuously. Server components render on the edge, off-screen content loads lazily, and images are sized per device to avoid oversized mobile downloads.

The bigger picture

UK homes produce a fifth of the country's carbon emissions

Traditional assessments are expensive and opaque. Making the analysis cheap and legible lowers the barrier to understanding — and then acting on — the improvements that actually matter.

Open source

Our carbon metrics use CO2.js from The Green Web Foundation. A published methodology is what makes the claims on this page checkable rather than decorative.

Sustainability is not a checkbox. This page is a live dashboard — as the stack changes, so do these numbers.

Ready to cut your home's carbon?

Start with the free check and find the improvements worth making first.

Check my home